APP_TIMER_V2 does not appear to generate timer interrupts

[RMV]

I am using SDK 17.1.0 with an nRF52832 design similar to the P10040.

I was previously using the original timer application and things were looking good.
I decided to upgrade to SDK 17.1.0 and am using the uart example under 'ble_peripherals' to boot strap my project.

Based on another project using SDK 17.1.0, I decided to move up to the newer version of the application timer.
I defined the symbols 'APP_TIMER_V2' and 'APP_TIMER_V2_RTC1_ENABLED' in the build.

My first question is:  I do not quite understand "why do I need to enable the RTC1"?  -- and when I try to build without using this symbol then I get a build error that there is no RTC instance defined in the build.

My next comment (based on observations) is that when I enable RTC1 then I notice that the processor enters the nrf_power_mgmt phase but never fires the timer handler that I expect to be triggered every 1 second. I have validated that
    a) the firmware has indeed called the 'app_timer_init()' function, 
    b) that this 1-second timer has been created as a 'repeat-mode' timer with a 1 second timeout with no errors, and
    c) that the timer has indeed been started.

I am sure that I am missing something here -- likely some kind of connection between the hardware RTC and the firmware handler but am still struggling to nail it down.

Cheers
RVM

[Karl Ylvisaker]

Hello again, RMV

Thank you for your patience with this.

RMV said:

FYI, I have noticed that when I reset the MCU (via the 5-pin SWD interface) and stop at the reset vector, the RUNSTATUS register in the WDT has the value '1' i.e. WDT is enabled. This does not match the product specification which states that this register should be '0' at reset.

I have reached out to the nRF5 SDK developers about this, and checked up on the internal issue that my colleague Susheel submitted in the other ticket, but there unfortunately does not seem to exist an estimate for when this fix might make it into an SDK release.

RMV said:

I really need to know ASAP whether or not this issue is going to be resolved, or there is some workaround outside of the SDK that I can use to get the LFCLK being recognized by the Applicatiion Timer?

I have discussed this with some colleagues and I think it comes down to that you will have to make sure that the LFCLK start task is triggered regardless of the WDT status.
You could for example do this through the HAL layer directly:

...
        nrf_clock_task_trigger(NRF_CLOCK_TASK_LFCLKSTART);
...

You can see the detailed explanation in this ticket.
Try this, and let me know if it resolves the issue you are seeing.

By the way, are you aware of the Errata 132 for the nRF52832? My colleague Håkon mentions its significance in the referenced ticket as well.

Best regards,
Karl

[RMV]

Karl Ylvisaker said:

I have discussed this with some colleagues and I think it comes down to that you will have to make sure that the LFCLK start task is triggered regardless of the WDT status.
You could for example do this through the HAL layer directly:

Fullscreen

I have tried this briefly and it seems to work -- time will tell how robust this is going to be.

I will mark this as 'answered'.  Please feel free to close this ticket.

Cheers

RVM

[Karl Ylvisaker]

Hello RVM,

I am glad to hear that this seems to resolve your issue - please do not hesitate to let us know if testing proves otherwise.
I will mark the ticket as 'closed' as requested, but you can re-open it at any time if it should be necessary.

Best regards,
Karl